Evaluation of Real Time Magnetic Resonance Imaging and Keypoint Detection Models for Assessing Range of Motion in Ankle Braces: A Study on Healthy Individuals

Introduction : Acute ankle sprains are prevalent injuries in physically active individuals. Functional support through the use of braces is a common approach to prevent excessive movements and reinjury to the affected ligaments. Recent advancements in imaging technology, particularly real-time magnetic resonance imaging (rtMRI), present new opportunities for evaluating physiological movements. This study aims to investigate the applicability of rtMRI in measuring the range of motion (ROM) of the talocrural joint while utilizing four different types of ankle braces. Methods: 45 dynamic rtMRI exams were performed in a cohort of 9 participants at the University Medical Center Göttingen with and without four commercially available brace models during plantar flexion, dorsiflexion, eversion and inversion. Two observers measured the angles between predefined anatomical keypoints at intervals of every ten images. In addition, a keypoint detection model (YOLO11) was trained and utilized as a third observer. Intraclass Correlation Coefficients (ICC) and the relative mobility restriction for each brace model were calculated. T-tests and linear mixed models were used for the analysis of differences between the brace models and observers. Results: All ankle braces significantly reduced the range of motion (ROM) in the sagittal plane by up to 38% compared to unrestricted movement. Sagittal measurements demonstrated a high level of interobserver agreement between two human observers and the keypoint detection model, with a consistency Intraclass Correlation Coefficient (ICC) of 0.92. This agreement was even higher in a subset of studies that adhered to predefined image quality standards, yielding a consistency ICC of 0.96. Coronal measurements showed low interobserver agreement presumably due to suboptimal image quality resulting from unwanted rotational movements, which influenced the visibility of the predefined anatomical landmarks. Conclusion: These findings suggests that rtMRI has the potential for accurate measurement of talocrural joint ROM in the context of evaluating ankle braces. However, imaging studies have to meet quality criteria to ensure high interobserver reliability. Furthermore, the implementation of keypoint detection algorithms can facilitate automated angle measurements, streamlining a time-consuming process for radiologists. Trial Registration: The study was retrospectively registered in the DRKS database on April 1 ^st , 2025 with the identifier DRKS00036544.